Multiple Myeloma Minimal Residual Disease Detection: Targeted Mass Spectrometry in Blood vs Next-Generation Sequencing in Bone Marrow

BACKGROUND: Minimal residual disease (MRD) status assessed on bone marrow aspirates is a major prognostic biomarker in multiple myeloma (MM). In this study we evaluated blood-based targeted mass spectrometry (MS-MRD) as a sensitive, minimally invasive alternative to measure MM disease activity. METHODS: Therapy response of 41 MM patients in the IFM-2009 clinical trial (NCT01191060) was assessed with MS-MRD on frozen sera and compared to routine state-of-the-art monoclonal protein (M-protein) diagnostics and next-generation sequencing (NGS-MRD) at 2 time points. RESULTS: In all 41 patients we were able to identify clonotypic M-protein-speciﬁc peptides and perform serum-based MS-MRD measurements. MS-MRD is signiﬁcantly more sensitive to detect M-protein compared to either electrophoretic M-protein diagnostics or serum free light chain analysis. The concordance between NGS-MRD and MS-MRD status in 81 paired bone marrow/sera samples was 79%. The 50% progression-free survival (PFS) was identical (49 months) for patients who were either NGS-positive or MS-positive directly after maintenance treatment. The 50% PFS was 69 and 89 months for NGS-negative and MS-negative patients, respectively. The longest 50% PFS (96 months) was observed in patients who were MRD-negative for both methods. MS-MRD relapse during maintenance treatment was signiﬁcantly correlated to poor PFS ( P < 0.0001). CONCLUSIONS: Our data indicate proof-of-principle that MS-MRD evaluation in blood is a feasible, patient friendly alternative to NGS-MRD assessed on bone marrow. Clinical validation of the prognostic value of MS-MRD and its complementary value in MRD-evaluation of patients with MM is warranted in an independent larger cohort.


Introduction
Multiple myeloma (MM) is the second most common hematological malignancy, characterized by the accumulation of clonal plasma cells, usually within the bone marrow (1). MM is an incurable disease, however the introduction of novel therapies has led to major improvements in clinical outcome (2). Routine monoclonal protein (M-protein) diagnostics play an important role in the screening and monitoring of MM patients (3). M-protein diagnostics are reliable, fast, and inexpensive, but not sensitive enough to detect low Mprotein concentrations (4,5). More than 50% of newly diagnosed patients reach a stringent complete response (6). However, most patients relapse despite achieving such deep responses, which stresses the need for new assays that can identify responses beyond conventionally defined stringent complete response.
Current methods to assess minimal residual disease (MRD) in MM patients focus on molecular and flow cytometric techniques performed on bone marrow aspirates (7,8). Generally, MRD-negativity is defined by the absence of clonal plasma cells on bone marrow aspirates with a minimum sensitivity of 1 myeloma cell in !10 5 nucleated cells. MRD assessment by multiparameter flow cytometry (MFC), allele-specific oligonucleotide (ASO)-quantitative PCR, or next-generation sequencing (NGS), can play an important role in the management of patients in complete remission. MRD status has evolved into a major prognostic factor (9)(10)(11). Moreover, MRD assessment can be applied to assess treatment effectiveness in clinical trials (12). Consequently, new International Myeloma Working Group (IMWG) consensus criteria for MRD assessment have been defined that reach beyond the detection of the present therapy response criteria (6). A disadvantage of MRD evaluation on bone marrow aspirates is the risk of non-representative sampling, resulting from heterogeneous dispersion of MM cells and possibly hemodilution (13). Extramedullary MM outgrowth may give negative results even after repetitive sampling. In addition, repetitive bone marrow biopsies are invasive to patients.
Blood-based mass spectrometry (MS) that targets M-protein-specific peptides has recently been introduced as a sensitive and minimally invasive alternative for MRD-evaluation performed on bone marrow (14)(15)(16). In this study we evaluated the feasibility of the MS-MRD assay in a cohort of 123 sera obtained from 41 MM patients in the IFM 2009 clinical study. For all patients, serum protein electrophoresis, immunofixation electrophoresis, free light chain analysis, NGS-MRD results, and a median of 51 months clinical follow-up were available (17). Blood-based MS-MRD performance was compared to conventional M-protein diagnostics and NGS-MRD evaluation performed on bone marrow.

PATIENT SAMPLES
All samples were collected from the IFM 2009 clinical study (ClinicalTrials.gov identifier NCT01191060) (17), after written informed consent, and clinical and genomic data were de-identified in accordance with the Declaration of Helsinki and approval for this study was provided by our Institutional Review Board (2018-4140). The IFM 2009 trial was a phase 3 study started in 2012 enrolling 700 patients under 66 years who were eligible for autologous stem cell transplantation (ASCT). Treatment started with 3 cycles of lenalidomide, bortezomib, and dexamethasone (RVD) followed by either ASCT or additional 5 RVD cycles. The disease evaluation was done before and after 1 year of lenalidomide maintenance. At these time points, the disease evaluation was assessed on both blood and bone marrow aspiration. Among the 700 patients included in the IFM 2009 trial, we selected patients for this study based on the following inclusion criteria: (a) availability of NGS-MRD data performed pre-maintenance and postmaintenance; (b) availability of RNA sequencing data; and (c) availability of serum samples at baseline, pre-maintenance and post-maintenance. In total 41 patients fulfilled all 3 criteria.

ROUTINE M-PROTEIN DIAGNOSTICS
M-protein diagnostics were performed in a centralized laboratory as routine diagnostic workup of the IFM 2009 clinical study. Serum protein electrophoresis (SPEP) was performed by capillary zone electrophoresis (Capillarys, Sebia) and immunofixation electrophoresis (IFE) was performed on the Hydrasys 2 system using the Hydragel 4 IF kit (Sebia). Serum free light chain (sFLC) measurements were performed using Freelite reagents (The Binding Site) on a BNII nephelometer (Siemens Healthcare), according to the manufacturer's protocol. Reference intervals for the sFLC ratio are 0.26 to 1.65.

MS-MRD
The M-protein sequences targeted in the MS-MRD approach were selected based on bioinformatics analysis of a 100,000-reads sample extracted from RNA sequencing data produced for the NCT01191060 trial (18). A contig was constructed based on targeted assembly of reads from highly expressed mRNA related to the M-protein.
Assembly was performed with the TASR software tool, directed by short seeding sequences from either conserved immunoglobulin regions, or from M-proteinspecific complementary-determining region (CDR)3 sequences that could be revealed by the MiXCR software package (19,20). Tryptic peptides from the Mprotein sequence were predicted in silico and assessed for containing mutations compared to an immunoglobulin germline reference (IMGT). Qualifying peptide targets were evaluated experimentally by targeted mass spectrometry analysis on the matching diagnostic serum sample in the cohort, as well as other serum samples as controls. For peptides yielding the strongest signal in the diagnostic serum sample while remaining negative in the controls, stable isotope-labelled peptides were synthesized (Pepscan). Samples were reduced with 10 mmol/L dithiotreitol, alkylated with iodoacetamide, and digested overnight with trypsin in 150 mL/L acetonitrile, 50 mmol/L ammonium bicarbonate. During LC-MS data acquisition, a synthetic peptide mixture was sampled at every 10th injection to validate performance of chromatography and mass spectrometry systems. All targeted LC-MS data were obtained after separation on a Pepmap Acclaim C18 column (75 mm x15 cm, 30 min gradient, 300 nL/min) and a Q-Exactive HF mass spectrometer (Thermo Fisher Scientific) at 120 000 resolution, 1x10 6 AGC target and 1.4 m/z isolation. The collision energy was optimized by analyzing a mixture of all stable isotope-labelled (SIL) peptides at different collision energies, ranging from 18 to 36 normalized collision energy. The collision energy resulting in the highest signal intensity was determined for each peptide and subsequently used for the measurements. The targeted masses, monitored transitions, and optimized collision energy of each peptide can be found in online Supplemental Table 1. All targeted LC-MS data were analyzed with Skyline; an rdotp score of >0.89 was used as the threshold for positivity (21). All MS-MRD negative test-results were confirmed by reinjecting these samples. Examples of MS-MRD positive and MS-MRD negative samples are shown in online Supplemental Fig. 1. Online Supplemental Fig. 2A shows a replicate analysis from 3 patients in whom all 3 sample time points were digested and analyzed in 3 independent experiments. Sample digest stability was tested in 9 samples in which sample digests were stored at 8 C for 12 weeks and re-analyzed on the same instrument. A linear fit of log-transformed data had a slope of 1.0091 and an R 2 of 0.9969 between fresh and 12-week-old samples (Supplemental Fig. 2, B). M-protein quantification was based on the average of the heavy and light chain clonotypic peptide, if applicable. For additional MS-MRD sensitivity of disease monitoring, all samples were analyzed a second time after 10 mL of serum was enriched for immunoglobulins using 100 mL of Melon Gel (Thermo Fisher Scientific) following the manufacturer's instructions. All MS-MRD analyses were performed blinded with respect to NGS-MRD results and survival data of the patients.

NGS-MRD
Collected bone marrow samples were frozen as dry pellets and stored at À80 C until analysis. MRD was performed as previously described (9). Briefly, the clonal rearrangements were identified using the NGS commercial Clonoseq kit for MRD assay (Adaptive Biotechnologies V R ). For MRD quantification, DNA was extracted from the bone marrow samples and amplified by polymerase chain reaction using immunoglobulin gene-specific primers; the amplified products were then sequenced. The sensitivity was 10 À6 , i.e., one malignant plasma cell within 1 000 000 bone marrow cells.

STATISTICAL ANALYSIS
GraphPad Prism 5.03 was used to perform statistical analyses and make the figures. The Kaplan-Meier method was used to estimate time-to-progression distributions, and statistical comparisons were done using log-rank (Mantel-Cox) tests.

IDENTIFICATION OF CLONOTYPIC PEPTIDES
Patient -specific clonotypic peptides were selected based on the assembled clone sequences derived from RNAseq datasets ( Fig. 1). For each heavy and light chain clone sequence in all 41 MM patients, multiple candidate clonotypic peptides could be identified from the immunoglobulin (Ig)-variable region. For the heavy chain on average 5.4 clonotypic peptides (range 2 to 7 peptides) and for the light chain on average 3.3 clonotypic peptides (range 2 to 5) were identified. Based on proteomics data, per M-protein 2 optimal clonotypic peptides were selected in terms of specificity and sensitivity in the mass spectrometer (Table 1). Clonotypic peptides were identified in sequences located at both the variable CDR and the more conserved framework regions, covering the full variable gene sequence (Table 1).

DIAGNOSTICS
In all 41 patients a baseline sample and 2 follow-up samples were measured using routine M-protein diagnostics (SPEP and IFE), free light chain analysis and MS-MRD. Of these 123 samples, 52 sera were negative using IFE (online Supplemental Table 2), sFLC ratio was normal in 69 sera (online Supplemental Table 3), and MS-MRD measured 17 MRD negative sera in total (Supplemental Tables 2 and 3). In 39 sera the results were MS positive whereas IFE results were negative. In 52 sera the MS result was positive whereas the sFLC ratio was normal. In all 17 MS-MRD negative samples, the IFE result was also negative and the sFLC ratio was normal. This indicates that routine M-protein diagnostics were never more sensitive than MS-MRD in any of the sera measured in this cohort.

METHOD COMPARISON MS-MRD WITH NGS-MRD
NGS-MRD was performed on 81 bone marrow aspirates in this cohort taken at both pre-maintenance and post-maintenance treatment. MS-MRD was performed on serum acquired at the same time points. Nineteen percent (15/81) of the samples were NGS-MRD negative and 21% (17/82) of the samples were MS-MRD negative. Overall concordance between both methods was 79% (Table 2).
Progression-free survival (PFS) analysis suggests that with both MRD methods MRD negativity is associated with longer PFS, however the longer PFS was not found statistically significant (P ¼ 0.14) which could be due to the relatively small cohort size (Fig. 2). The 50% PFS was 49 months for patients who were NGS-positive directly after maintenance treatment ( Table 3). The 50% PFS of MS-positive patients was identical (49 months). The 50% PFS was 69 and 89 months for NGS-negative and MS-negative patients, respectively. The longest 50% PFS (96 months) was observed in patients who were MRD-negative for both methods. For quantitative MS-MRD measurements, SIL clonotypic peptides with a known concentration were synthesized for each patient. To quantify the M-protein, the quantifications of the heavy and light chain clonotypic peptides were averaged. Small differences between heavy and light chain clonotypic peptides were observed, but they showed similar trends between the 3 measured time points (online Supplemental Fig. 3). To further increase the sensitivity of the MS-MRD assay, a pre-analytical antibody purification step using Melon Gel was introduced (Fig. 1, B). The M-protein could be quantified down to approximately1 mg/L with a sufficient signal-to-noise ratio (online Supplemental Fig.  4). Re-analysis of all 82 MRD samples with SIL calibrators and Ig-enrichment reduced the number of MS-MRD-negative samples from 21% to 11% (9/82 samples).

DISEASE MONITORING USING MS-MRD
MS-MRD quantification by addition of the SIL calibrators allows for the first time monitoring of M-protein concentrations in patients who achieve stringent complete remission. We observed that the M-protein concentration in 35 out of 41 patients (85%) decreased during maintenance treatment; in the remaining 6 patients (15%) we observed an increased M-protein concentration during maintenance treatment (Fig. 3,  A). MS-MRD relapse during maintenance treatment was significantly correlated to poor PFS (Fig. 3, B, P < 0.0001).

Discussion
Driven by the evolving framework of more effective multidrug treatment regimes, new methods have been developed to detect and quantify MRD in MM patients   (22). Consequently, new IMWG consensus criteria for MRD assessment have been defined that reach beyond the detection of the present therapy response criteria (6). In this study we evaluated MS-MRD performance in a cohort of 41 MM patients from the IFM 2009 clinical study. Blood-based MS-MRD performance was compared to conventional M-protein diagnostics and NGS-MRD evaluation performed on matching bone marrow samples. MS-MRD testing necessitates identification of Mprotein-specific clonotypic peptides in each patient. In our recent in silico study performed on a cohort of 609 patients, we demonstrated that multiple clonotypic peptides within the variable part of the M-protein could be identified from RNA sequence data in all MM patients (23). In this study we confirmed that on average 8 MSsuitable clonotypic peptides could be identified in each patient. For a robust and confident M-protein quantitation by the MS-MRD assay, we selected a clonotypic M-protein peptide from both the heavy and light chain based on highest MS intensity for optimal sensitivity and specificity. When both clonotypic peptides are measured, it becomes feasible to also monitor disease activity in patients with free light chain escape (24). In practice, we did not observe this phenomenon in the current cohort. Three out of 41 patients had a light chain-only MM at study entry and for these patients we selected both peptides from the clonal light chain. The 150-bp paired-end sequencing on HiSeqX10 genome analyzers provided full coverage of the Ig-clone. That is an advantage over other sequencing approaches that limit information to a segment containing the CDR3 due to primer choice in a PCR amplification step. It allowed us to include clonotypic peptides from the entire variable region of the M-protein.
For ease of use, we initially performed the MS-MRD analyses directly on serum without enrichment of immunoglobulins and without addition of SIL calibrators to obtain qualitative MS-MRD data. MS-MRD performed on 123 sera was compared to routine electrophoretic M-protein diagnostics and FLC analyses that were performed on the same sera. In total 52 sera tested IFE negative, 69 sera had a normal sFLC ratio, and 17 sera were MS-MRD negative. All sera that were either IFE positive (n ¼ 64) or had an abnormal sFLC ratio (54) were also found MS-MRD positive. This indicates that MS-MRD is more sensitive than IFE and sFLC analysis in all tested samples.
In 2 recent method comparisons between NGS and MFC MRD-evaluation performed on the same bone Table 3. Progression-free survival (PFS) and correlation to MRD-status.    marrow aspirates, concordance rates of 68% (25) and 93% (26) were described. The overall concordance rate between NGS-MRD and MS-MRD in this study was 79%. That is high, considering that the two methods measure different biomarkers, as NGS measures DNA reads from myeloma cells in the bone marrow and MS measures M-protein in the blood. Our data are in line with the 63% concordance between NGS-MRD and clonotypic peptide MS analysis on serum that was recently reported by Derman et al. (15). Compared to NGS-MRD, their blood-based MRD assay was at least as sensitive to detect MRD and appeared to be a superior predictor of PFS (15). Also Martins et al. recently showed that targeted proteomics to detect clonotypic peptides in the serum of patients with MM is possible, even in patients who were tested negative with MFC MRD evaluation performed on bone marrow (16). For patients with extramedullary myeloma disease spreading, blood-based MS-MRD will likely be more reliable. In contrast, in patients with non-secreting MM NGS performed on bone marrow will likely be more reliable. Another confounding factor for the evaluation of MS-MRD results, is the half-life of M-proteins in the blood which is on average 21 days for IgG and 10 days for IgA. This causes a delay between lysis of clonal plasma cells and the decrease in M-protein. In individual patients it is difficult to prove the exact reason for a discordant result between NGS and MS. The complementary nature of both MRD analyses was further stressed by the fact that longest PFS was observed in patients who were MRD negative for both NGS and MS.
In this study we showed that the sensitivity of the MS-MRD analysis can be further increased using preanalytical Ig-enrichment procedures and the introduction of SIL calibrators. Pre-analytical Ig-enrichment and addition of SIL calibrators increases the confidence with which low-concentration M-proteins are measured. As a consequence, an increasing number of patients test MS-MRD positive. In this cohort, only 9 out of 82 MRD sera were evaluated as MRD positive. In order to operate as a biomarker with the same prognostic value as NGS-MRD, this would necessitate the introduction of an MS-MRD cutoff value. The MS-MRD cutoff value that corresponds to the current MRD cutoff (1 myeloma cell in !10 5 nucleated cells), must be assessed in a future study in a large independent cohort. The lower limit of our MS-MRD quantification method is estimated to be 1 mg/L M-protein (27).
While an MRD-negative treatment response is favorable, sustained MRD negativity is likely to be associated with even better prognosis. Sequential MRD analysis further has the potential to detect relapses early, before overt biochemical or clinical relapse (28). MS-MRD quantification was possible by addition of SIL calibrators, which allowed for the first time monitoring of M-protein concentrations in patients who achieve stringent complete remission. In this study we showed that MS-MRD relapse during maintenance treatment was significantly correlated to poor PFS. It is important to note that it takes 4 to 5 weeks to develop and validate a patient-specific MS-MRD assay. The MS-MRD measurements for subsequent monitoring of a patient can be processed within 2 days, which makes it very suitable for dynamic MRD monitoring. For clinical implementation of MS-MRD this long turnaround time for the first MS-MRD measurement is likely not problematic since MRD evaluation is not relevant within the first half year after the start of therapy. Additionally, Martins et al. have shown that stable isotope-labelled control-Ig might be able to replace the most time-consuming element, namely the synthesis and validation of the SIL calibrators (16).

Conclusion
Our data indicate proof-of-principle that MS-MRD evaluation in blood is a feasible, patient friendly alternative to NGS-MRD assessed on bone marrow. MRD status obtained in bone marrow provides information that cannot be achieved by MS, such as clone evolution and bone marrow reconstitution. As such, we anticipate that in the future MS will not replace existing MRD tests on bone marrow, but will have clinical value as a companion method especially for minimally invasive, longitudinal monitoring of MRD in blood. Further research is warranted to specify the exact role of MS-MRD analyses in the management of MM patients in larger cohorts.

Supplemental Material
Supplemental material is available at Clinical Chemistry online.
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 4 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved. Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, preparation of manuscript, or final approval of manuscript.